# Claude Code game ban: Anthropic now blocks recreating existing video games

> Two sentences added to Claude Code on Opus 5.5 tell it to refuse clones of specific games, and developers say the block reaches further.

- Publisher: NextDiff (https://nextdiff.com/)
- Section: Policy
- Author: NextDiff Editorial
- Published: 2026-11-01T09:48:01.002Z
- Updated: 2026-11-01T09:48:01.002Z
- Canonical URL: https://nextdiff.com/posts/claude-code-game-ban-anthropic-blocks-game-clones/
- Topics: ai-gaming, video-games, agents, llms

## In short

Anthropic added two sentences to Claude Code on Opus 5.5, pushed on 27 October 2026, telling the agent not to recreate a specific existing video game even if renamed. Original and genre games still work. Anthropic says cloning would breach its IP rules; users report wider refusals, which Anthropic has not confirmed.

## Key takeaways

- Since a 27 October rules update, Claude Code on Opus 5.5 is told not to build a recreation of a specific existing game, even if it is renamed.
- Original games, genre games and games built from your own designs and assets are still allowed, Anthropic’s Addy Osmani said.
- Users report wider refusals, including generic reverse-engineering and troubleshooting tasks; Anthropic has confirmed only the game rule.
- There is no setting to turn the rule off, and Anthropic has not said whether other Claude models carry it.

Anthropic has quietly changed what Claude Code will build. Since a rules update pushed on 27 October, the coding agent running on Opus 5.5 is told not to recreate a specific existing video game, "even if it is renamed or called original." Anthropic confirmed the change after an X user spotted it. Here is what the Claude Code game rule actually says, what it does not block, and why developers are split.

![The two sentences Anthropic added to Claude Code: do build original games; do not recreate a specific existing game](/media/41745701292d4bed819d.jpg)
*Graphic: NextDiff, from the rule text reported by Game Developer*

## What changed in Claude Code?

An X user called Chasm found a new instruction in Claude Code's rules, listed under the internal name "tengu_heron_brook". [Game Developer](https://www.gamedeveloper.com/programming/anthropic-appears-to-have-ordered-ai-tool-claude-to-stop-remaking-video-games) first reported it on 30 October. The rule has two halves:

- **Allowed:** "Do build original games, games in a genre, and games from the user's own designs and assets."
- **Not allowed:** "Do not build a recreation of a specific existing game, even if it is renamed or called original."

Anthropic did not publish an announcement. A spokesperson pointed Game Developer to a post on X by Addy Osmani, a member of Anthropic's technical staff. Osmani said the company added two sentences to Claude Code's prompt on Opus 5.5 to "steer it away from recreating someone else's specific game," because using Claude to infringe IP breaks its usage policy. He added: "Original games, genre games, and games built from your own designs and assets are meant to keep working."

## Is Claude Code banning game development?

No. The rule targets clones, not games. You can still ask Claude Code for a platformer, a roguelike, a farming sim or a match-three puzzle, because those are genres. You can still build from your own art, levels and design documents. What it is told to refuse is "make me Stardew Valley" or the same request with the name changed.

The line is blurry in practice. As [tbreak](https://tbreak.com/claude-code-stops-recreating-video-games/) notes, a fan project, an homage or a game that is a clone in all but name could land on either side. Chasm reported that there is no setting to switch the rule off, and Anthropic has not said whether models other than Opus 5.5 in Claude Code carry the same instruction.

![Claude Code welcome screen in a terminal](/media/8df81d14ca3d49f68367.jpg)
*Image: Anthropic*

## What are developers complaining about?

The complaints go further than game clones. [Tom's Hardware](https://www.tomshardware.com/tech-industry/artificial-intelligence/anthropics-ai-model-is-now-blocking-attempts-at-video-game-recreation-and-cloning-users-also-claim-non-malicious-generic-reverse-engineering-tasks-are-blocked) reports user claims that harmless, generic reverse-engineering work and basic troubleshooting tasks are now being refused too, and Wccftech says developers report legitimate projects being blocked. Anthropic has only confirmed the game-recreation sentences, so treat the wider blocking as user reports for now. Osmani said Anthropic will "continue to iterate to ensure we're striking the right balance."

On Hacker News, a separate thread about Anthropic blocking accounts for work related to training larger models drew more than 300 points. That claim rests on one user's account, and we found no Anthropic statement about it. The written rule is clear, though: Anthropic's [Usage Policy](https://www.anthropic.com/legal/aup) bars using Claude's outputs to train an AI model without permission, and it lists infringing a third party's intellectual property as a violation.

## Why is Anthropic doing this?

Cloning has become a real problem for small studios. AI coding agents make it far quicker to copy a successful indie game. According to [PC Gamer](https://www.pcgamer.com/software/ai/valve-admits-steam-has-an-ai-slop-and-cloning-problem-but-says-we-dont-want-to-be-gatekeepers-telling-developers-which-tools-they-should-or-shouldnt-use/), Valve admitted this week that Steam has an AI slop and cloning problem, but said it does not want to police which tools developers use. Some indie teams have cancelled public playtests over fears that their games would be cloned before launch.

![Crowds at the Independent Games Festival pavilion at GDC 2013](/media/fe1363b323304da88b3e.jpg)
*Image: Official GDC (CC BY 2.0), via Wikimedia Commons*

## Why it matters

This is a policy story as much as a product one. With Valve stepping back, the AI vendor is the one drawing the line, and it does so through a hidden system prompt rather than a law, a store rule or a published changelog. Developers only learned of it because someone read the prompt. That raises questions every team building on AI agents should ask: which rules ship in the model's instructions, who gets told when they change, and what happens to legitimate work, such as remakes you hold the rights to, that a two-sentence rule cannot tell apart from piracy.

It also shows where AI coding is heading. As agents like Claude Code take on whole projects (the trend we described in [What is vibe coding?](/posts/what-is-vibe-coding/)), vendors will add more guardrails of this kind. Anthropic, which filed for its [IPO last week](/posts/anthropic-ipo-s1-revenue-loss-ticker/), has every reason to avoid being the tool behind a wave of copied games.

## What's next

Anthropic says it will keep tuning the rule, and the usage policy notes that hitting a block does not by itself mean you broke the rules. If Claude Code refuses an original project, rephrase the request around your own design and assets. Watch for whether the rule spreads to other Claude models, and whether OpenAI and Google add similar limits to their coding agents.

## Questions readers ask

### Can Claude Code still make games?

Yes. Anthropic says original games, games in a genre and games built from your own designs and assets are meant to keep working. Only recreations of a specific existing game are refused.

### Why did Anthropic block game cloning in Claude Code?

Anthropic’s Addy Osmani said the two sentences were added because using Claude to infringe intellectual property is against Anthropic’s usage policy.

### Can you turn off the Claude Code game restriction?

No. The rule is part of Claude Code’s prompt on Opus 5.5, and the X user who found it reported there is no setting to disable it.

### Does Claude Code block reverse engineering?

Some users say generic reverse-engineering and troubleshooting tasks are being refused, according to Tom’s Hardware. Anthropic has only confirmed the rule about recreating existing games.

### When did the Claude Code game rule start?

The rule was pushed to Claude Code around 27 October 2026, spotted on 30 October and confirmed by Anthropic the same day, according to Game Developer.

## Sources

- [Game Developer — Anthropic orders AI tool Claude to stop remaking video games](https://www.gamedeveloper.com/programming/anthropic-appears-to-have-ordered-ai-tool-claude-to-stop-remaking-video-games)
- [Tom’s Hardware — Anthropic’s Claude AI blocks video game cloning and reverse-engineering](https://www.tomshardware.com/tech-industry/artificial-intelligence/anthropics-ai-model-is-now-blocking-attempts-at-video-game-recreation-and-cloning-users-also-claim-non-malicious-generic-reverse-engineering-tasks-are-blocked)
- [tbreak — Claude Code blocks recreating existing video games](https://tbreak.com/claude-code-stops-recreating-video-games/)
- [Anthropic — Usage Policy](https://www.anthropic.com/legal/aup)
- [Chasm on X — the Claude Code rule change](https://x.com/chasmmmmmmmmmmm/status/2116062991686721968)
- [PC Gamer — Valve admits Steam has an AI slop and cloning problem](https://www.pcgamer.com/software/ai/valve-admits-steam-has-an-ai-slop-and-cloning-problem-but-says-we-dont-want-to-be-gatekeepers-telling-developers-which-tools-they-should-or-shouldnt-use/)